Short version:
The MultiAir/UniAir solenoids are
very low-resistance coils that are driven from 12 V with a
high-frequency PWM carrier (~31.5 kHz), and the ECU shapes the
coil current (not just the voltage) in a
bias → peak → hold profile. On top of that, the ECU turns the
whole PWM stream on and off at specific crank angles to get the required valve lift curve.
Below is how it all hangs together.
1. Electrical characteristics of the MultiAir solenoids
Coil resistance and current
Workshop docs for the 1.4 MultiAir give a typical solenoid resistance of about
0.25–0.30 Ω.(
forum.alfaholicy.org)
The Racewicz paper you’re quoting measured that each solenoid needs roughly
8 A of coil current to switch reliably.(
kones.eu)
Because ( I = V/R ), a 0.25–0.30 Ω coil on a 12 V battery would try to pull
40–50 A if you just slammed 12 V DC across it, which would cook things very quickly. That’s why the ECU uses PWM and a current-controlled driver rather than a simple on/off 12 V feed.
2. The PWM carrier from the ECU
From the oscilloscope captures in the Racewicz paper (the bit you quoted
kones.eu)
- Amplitude: ~12 V
- Frequency: ~31.65 kHz
- Duty cycle (average): ~18.5 %
- Equivalent DC voltage: ~2.2 V (12 V × 0.185)
So electrically you can think of it as a
12 V square wave at 31.5 kHz, duty-cycled so that the
average voltage seen by the coil is ~2–3 V, which is enough to get 8–10 A through a ~0.25–0.30 Ω winding once inductance and wiring are taken into account.
Why such a high frequency?
- With such low coil resistance, high frequency + low duty lets them limit the average current to a sane value.
- 30 kHz is well above the audible range, so the coil doesn’t “buzz”.
- The coil inductance smooths the current, so at 30 kHz the current ripple is small and the ECU can treat it as essentially DC.
Racewicz’s external controller literally
multiplies an “enable” signal by this constant 31.65 kHz carrier (logic gate), then feeds a MOSFET that sinks current from the solenoid.(
kones.eu) That’s also effectively what the OEM ECU does internally.
3. Current profile: bias, peak, hold
Fiat’s own eLearn documentation for the Magneti Marelli 8GMF (the MultiAir ECU) describes the solenoid drive as a
special current profile with three phases).
- Bias phase
- Average current ≈ 4 A
- Keeps the solenoid “ready” and overcomes static friction / oil forces so response is fast when the main command comes.
- Peak phase
- Current quickly ramped up to ≈ 11 A max
- Ensures very fast armature movement so the hydraulic valve actually shifts within a fraction of a millisecond.
- Hold phase
- Current reduced and held around 5 A
- Enough to keep the solenoid in the commanded position without wasting power or overheating.
The 31.5 kHz PWM duty cycle is what the ECU varies to generate those three current levels. Internally it’s a
current-controlled, peak-and-hold driver: it monitors coil current (there are specific DTCs for “UniAir electrovalve drive current feedback”).(
forum.alfaholicy.org)
So, when you see “18.5 % duty = 2.2 V” in that paper, that’s basically the
average drive during one of those current phases, not the whole story of what the ECU can do.
4. Timing within the engine cycle
Now the more interesting part:
when those solenoids are energized.
Mechanically, MultiAir works like this: a hydraulic piston follows the
intake cam lobe and pushes oil into the valve actuator. A
normally-open hydraulic solenoid controls whether that oil pressure is transmitted to the valve or dumped back to a small chamber.(
Stellantis Media)
- Solenoid closed (electrically energized): hydraulic path is “solid”; valve follows the cam (full mechanical lift).
- Solenoid open (de-energised): pressure is vented; the valve decouples and closes under its spring.
So the ECU doesn’t “meter lift” directly. Instead, for
each cam lobe pass (each intake event for that cylinder), it decides:
At what crank angle do I turn the solenoid on (close it)?
At what crank angle do I turn it off (open it again)?
These two angles define the valve profile, within the envelope of the mechanical cam.
Typical time scales
Take the example in the Racewicz measurements:
1500 rpm.(
kones.eu)
- 1500 rpm crank → 25 rev/s → one crank rev ≈ 40 ms.
- Camshaft runs at half speed → one cam rev ≈ 80 ms.
- The intake event occupies maybe ~120° of cam rotation (≈240° crank), so ~25–30 ms of real time.
Within that ~25 ms window, the ECU will:
- Start the solenoid drive (bias → peak → hold) somewhere around the opening flank of the mechanical cam profile.
- Maintain or release it depending on the desired mode (full lift, early closing, late opening, multi-lift).
The actual PWM carrier at 31.5 kHz has a period of ~
32 µs, so during one valve event (~25 ms) you have hundreds of PWM periods. The
on/off “envelope” of those periods is what corresponds to the timing of the valve event.
5. How different valve modes map to solenoid timing
Using the standard MultiAir modes (from Fiat/Schaeffler technical docs and training material
kones.eu)
- Full Lift (high load, high rpm)
- Solenoid is energized (closed) for the entire cam lobe event.
- Hydraulic link is solid → valve follows cam fully.
- Early Intake Valve Closing – EIVC (part-load torque, low–mid rpm)
- Solenoid starts closed, then the ECU de-energizes (opens) it before the cam lobe ends.
- Oil is dumped early → valve closes earlier than the cam would, trapping a controlled air mass without throttle losses.
- Late Intake Valve Opening – LIVO (idle / very light load)
- Solenoid is open at the start of the lobe, then the ECU energizes it later in the lift event.
- Valve opens late and with lower effective lift, shortening the event and improving mixture speed at low load.
- Multi-lift (two smaller valve events in one cam lobe)
- Solenoid is toggled closed → open → closed within a single cam event.
- This gives two smaller valve openings, useful for emissions and combustion control in some operating points.
The oscilloscope plots in the paper you cited (Fig. 8a vs 8b) are exactly this: at the same engine speed, the
solenoid “on” window is different for full-load vs minimal-load, changing when the intake valves open and shut even though the cam lobe itself is fixed.
6. Putting it all together
So if you imagine the command to one MultiAir solenoid for a single intake stroke, electrically it looks like:
- Enable window(e.g. 10–25 ms long at 1500 rpm)
- ECU decides start & end time based on crank angle, load, etc.
- Within that window: current profile
- A short bias period (~4 A), then a peak (~11 A), then hold (~5 A) to keep the spool shifted.
- Carrier PWM
- The whole thing is implemented as a 12 V, 31.5 kHz square wave whose duty cycle the ECU continuously adjusts to get the required coil current.(kones.eu)
Outside the enable window, the PWM duty goes to zero, the current decays, the electromagnetic solenoid de-energizes and its return spring opens the hydraulic path, and the valve is allowed to close under its spring force.
